J Neurophysiol 100: 1372–1383, 2008.
First published July 2, 2008; doi:10.1152/jn.00023.2008.
Metamorphosis-Induced Changes in the Coupling of Spinal Thoraco-Lumbar
Motor Outputs During Swimming in Xenopus laevis
Anna Beyeler, Charles Métais, Denis Combes, John Simmers, and Didier Le Ray
Université de Bordeaux; Centre National de la Recherche Scientifique, Laboratoire Mouvement Adaptation Cognition (UMR 5227),
Bordeaux; France
Submitted 9 January 2008; accepted in final form 12 July 2008
INTRODUCTION
Many behavioral tasks in invertebrates and vertebrates alike
are driven by primary motor programs that arise from neuronal
network assemblages within the CNS. Vertebrate locomotion
is one of the best understood of such functions, the basic motor
program for which being generated by local spinal networks
that are controlled by higher centers in the brain stem and are
continually shaped by sensory information from the periphery
(Rossignol et al. 2006). Over the last 30 years, many studies on
a variety of animal species have analyzed the functional organization of spinal and supraspinal structures involved in the
orchestration of rhythmic locomotor movements (for reviews,
see Cazalets and Bertrand 2000; Grillner 2006; Kiehn 2006). In
quadrupedal vertebrates, the locomotor programs for anterior
Address for reprint requests and other correspondence: D. Le Ray, Université de Bordeaux; Centre National de la Recherche Scientifique, Laboratoire
Mouvement Adaptation Cognition (UMR 5227), 146 Rue Léo Saignat, 33076
Bordeaux; France (E-mail: [email protected]).
1372
and posterior limb movements are elaborated by so-called
“central pattern generators” (CPGs) (for reviews, see Grillner
1981; Kiehn 2006) that are located in the cervical (Ballion
et al. 2001; Juvin et al. 2005) and lumbar spinal enlargements,
respectively (Cazalets et al. 1995; Langlet et al. 2005). For
meaningful locomotory behavior, however, musculature in
other body regions such as the trunk and neck must be
coordinated with limb movements to ensure postural stability
during body displacements. Rhythmic activation of axial trunk
muscles in time with appendicular locomotor activity has been
reported in rat (Falgairolle and Cazalets 2007; Gramsbergen
et al. 1999), cat (Koehler et al. 1984; Zomlefer et al. 1984), and
humans (Thorstensson et al. 1982). However, although earlier
experiments on decerebrate paralyzed cat suggested a propriospinal origin of coordinated limb-trunk muscle contractions
(Koehler et al. 1984), the neural basis of such coupling and the
extent to which it is governed by rhythmogenic spinal circuitry
remains largely unknown (Falgairolle and Cazalets 2007).
Similarly, little is known of the changing functional relationship that must occur between locomotory and postural control
systems during postnatal development.
Metamorphosis from tadpole to frog in anuran amphibians
like Xenopus laevis constitutes one of the most striking developmental transformations in biology, involving fundamental
alterations in virtually all of the animal’s physiological systems
and body structure (for a review, see Shi 2000). One of the
most dramatic changes occurs in the biomechanical apparatus,
whereby the tail is resorbed and new limbs are formed as the
organism changes its mode of locomotion from tail-based
undulations in larvae to limb-based propulsion in the adult. In
premetamorphic tadpoles, body displacement is driven by
waves of bilaterally alternating muscle contractions that are
directed rostrocaudally along the body axis, whereas in postmetamorphic juveniles, bilaterally synchronous hindlimb kicking is principally used to propel the animal forward. Previous
studies strongly suggested that this metamorphosis-induced
transition in locomotor strategy results principally from the
gradual emergence of a lumbar CPG specifically dedicated to
the control of the newly developed hindlimbs (Combes et al.
2004; Rauscent et al. 2007). However, the functional destiny of
original axial motor circuitry that persists in the adult spinal
cord remains to be determined. Although the caudal spinal cord
segments that control axial movements in tadpoles disappear
with tail resorption after metamorphic climax, the segments
above the lumbar enlargement are preserved in adulthood and
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked “advertisement”
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
0022-3077/08 $8.00 Copyright © 2008 The American Physiological Society
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
Beyeler A, Métais C, Combes D, Simmers J, Le Ray D. Metamorphosis-induced changes in the coupling of spinal thoraco-lumbar motor
outputs during swimming in Xenopus laevis. J Neurophysiol 100:
1372–1383, 2008. First published July 2, 2008; doi:10.1152/jn.00023.2008.
Anuran metamorphosis includes a complete remodeling of the animal’s
biomechanical apparatus, requiring a corresponding functional reorganization of underlying central neural circuitry. This involves changes that
must occur in the coordination between the motor outputs of different
spinal segments to harmonize locomotor and postural functions as the
limbs grow and the tail regresses. In premetamorphic Xenopus laevis
tadpoles, axial motor output drives rostrocaudally propagating segmental
myotomal contractions that generate propulsive body undulations. During metamorphosis, the anterior axial musculature of the tadpole progressively evolves into dorsal muscles in the postmetamorphic froglet in
which some of these back muscles lose their implicit locomotor function
to serve exclusively in postural control in the adult. To understand how
locomotor and postural systems interact during locomotion in juvenile
Xenopus, we have investigated the coordination between postural back
and hindlimb muscle activity during free forward swimming. Axial/
dorsal muscles, which contract in bilateral alternation during undulatory
swimming in premetamorphic tadpoles, change their left-right coordination to become activated in phase with bilaterally synchronous hindlimb
extensions in locomoting juveniles. Based on in vitro electrophysiological experiments as well as specific spinal lesions in vivo, a spinal cord
region was delimited in which propriospinal interactions are directly
responsible for the coordination between leg and back muscle contractions. Our findings therefore indicate that dynamic postural adjustments
during adult Xenopus locomotion are mediated by local intraspinal
pathways through which the lumbar generator for hindlimb propulsive
kicking provides caudorostral commands to thoracic spinal circuitry
controlling the dorsal trunk musculature.
LUMBO-THORACIC COUPLING DURING AMPHIBIAN LOCOMOTION
wires was performed under light anesthesia after a small incision had
been made in the overlying skin, then the electrodes were fixed to the
muscle surface with a spot of Vetbond 3M adhesive (World Precision
Instruments). The skin was then replaced and attached with glue.
EMG signals were directed to a computer through a CED Micro 1401
interface (Cambridge Electronic Design) for storage and later analysis
using Spike 2 (CED) software.
Surgery for spinal lesions in juveniles
Juvenile frogs were individually anesthetized with freshly dissolved
tricaine methanesulfonate (MS 222, 50 mg/l, Sigma-Aldrich) and
positioned dorsal side up in a Petri dish filled with frog saline (which
contained, in mM: 112 NaCl, 2 KCl, 20 NaHCO3, 2.8 CaCl2, 1
MgCl2, and 17 glucose). After an incision was made in the dorsal skin,
muscles were carefully removed, and the underlying vertebrae were
opened dorsally along the thoracic and/or lumbar cord regions. To
separate the thoracic spinal cord from other regions of the nervous
system, three types of lesion were made separately or in combination:
a transection at the level of the last cervical cord segment, a longitudinal (sagittal) lesion that extended through only the three thoracic
segments, and a hemi- or whole-cord transection between the last
thoracic and first lumbar segments.
After lesioning, the spinal cord was cleaned and covered with a
small piece of gauze compress soaked in saline. The skin was then
replaced and attached with Vetbond glue. Animals were allowed 24 h
to recover from anesthesia and surgery before EMG recordings
commenced.
The extent of spinal lesions was verified after experimental recordings, and only data from animals with appropriate lesions were
analyzed in this study. Control experiments on operated juvenile
Xenopus but with the spinal cord remaining intact (n ⫽ 4) showed no
significant changes in either their patterns of free swimming or the
coordination between dorsalis trunci and plantaris longus muscle
activity (data not shown). EMG recordings in pre- and prometamorphic tadpoles were restricted to rostral axial muscles at a level that
corresponded to the adult thoracic cord region.
Retrograde staining of dorsalis motoneurons
METHODS
Experiments were performed on four pre- and three pro-metamorphic tadpoles (stages 54 –57 and to 60 – 61, respectively) (Nieuwkoop
and Faber 1956) and 50 postmetamorphic juvenile adults (stage 66,
⬍1 mo after metamorphosis) of the South African clawed toad, X.
laevis, bred from an in-house laboratory colony. Tadpoles were raised
until near stage 50 at 20°C and then kept at room temperature
throughout metamorphosis. EMG recordings were performed on animals at room temperature in separate aquaria, while in vitro experiments were performed at 16 –18°C. All procedures were in keeping
with the animal care guidelines of the Bordeaux University and the
CNRS.
In vivo EMG recordings
EMG activity of rostral axial musculature in larvae and back and
leg muscles in juveniles was recorded using pairs of 50-␮m insulated
wire electrodes connected through a grounded cable to a differential
AC amplifier (Model 1700, AM-System). Axial muscles were recorded at the level of the fifth segment of the tadpole spinal cord,
corresponding to the segment that commands the future back muscles
in postmetamorphic juveniles. In the latter, EMG recordings were
made on both sides of the trunk from the third myomere of the dorsalis
trunci, which is located dorsally along the vertebral column at middistance between the fore- and hindlimbs (Vallois 1922), and from the
ankle extensor plantaris longus (often called gastrocnemius in frogs)
(Peters 2005) of both hindlimbs. The implantation of EMG recording
J Neurophysiol • VOL
The somata of motoneurons innervating the dorsalis trunci muscles
were located in the spinal cord by means of retrograde axonal staining
from target muscles in vivo. Four postmetamorphic animals were
anesthetized in MS 222 and placed on ice in a Petri dish. A short
incision was made in the skin along the dorsal midline to gain access
to the back muscles. Small crystals of two fluorescent dyes (alexa
fluor 488 and 546 coupled to dextran 10,000) were inserted separately
with a fine pin into the left and right third myomeres of dorsalis trunci.
The skin was then resealed with glue, and animals were placed in
separate aquaria for recovery. After 3 days to allow retrograde dye
migration across the neuromuscular junction and along motoneuron
axons to the CNS, animals were killed and their spinal cords dissected
out and fixed overnight at 4°C in 4% paraformaldehyde in phosphate
buffer 0.1 M, pH ⫽ 7.4. After dehydration, the cords were cleared in
glycerol and examined with a fluorescent microscope (Leica-DMRB)
equipped with a CCD camera (ORCA-AG Hamamatsu). Images were
acquired using Simple PCI software (Compix).
Isolated brain stem-spinal cord preparations
Dissection and recording procedures were similar to those previously described by Combes et al. (2004). The skin and cranial
cartilage of juvenile frogs were opened under deep anesthesia with
tricaine methanesulfonate, and the forebrain was removed above the
mesencephalon. The spinal cord and remaining brain stem were
dissected out together with the thoracic ventral roots and the nerve
branches that innervate the hindlimb tibialis anterioris and the plan-
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
must also adapt to the new body format. To what extent do
developmental changes in the organization of segmental networks relate to the animal’s needs for postural control during
swimming? More specifically, what coordinating processes
enable certain axial muscles, which are directly engaged in
body propulsion in the premetamorphic larvae, to assume a
dynamic postural function after their transformation into nonlocomotory back muscles in postmetamorphic adults?
To begin to address these questions, we have analyzed the
bilateral coordination of axial/back muscle activity in freely
behaving, premetamorphic Xenopus tadpoles and, together
with the locomotor activity of hindlimb muscles, in postmetamorphic young adults. In the latter, moreover, simultaneous
electromyographic (EMG) recordings were made from back
and limb muscles on both sides of the body during free
straight-ahead swimming, then a series of spinal cord lesions
was performed to better understand the neural origins of the
temporal relationships between activity in these muscle sets in
vivo. We show that the rostrocaudal recruitment sequence and
left-right alternation of thoracic axial muscles in the tadpole is
replaced during metamorphosis by a different adult coordination pattern in which back and leg muscles contract synchronously during swimming, in a bilaterally in-phase pattern.
Based on in vivo lesions and in vitro experiments on isolated
brain stem-spinal cord preparations, it appears that the lumbar
CPG for hindlimb locomotion is also directly responsible for
driving thoracic motor output to dorsal muscles in postmetamorphic Xenopus. Given the developmental switch in function
of the more rostral axial muscles in larvae to nonpropulsive
trunk musculature in the adult frog, the Xenopus model should
help to provide new insights into the dynamic interactions
between locomotory and postural control systems in general.
Part of this work has been presented previously in abstract
form (Beyeler et al. 2007).
1373
1374
A. BEYELER, C. MÉTAIS, D. COMBES, J. SIMMERS, AND D. LE RAY
Data analysis
All analyses of electrical recordings were performed using homemade scripts running under Spike 2 (Spike 2 language, CED). Raw
signals were first integrated, and only activity sequences obtained
during episodes of rectilinear forward swimming were analyzed
further. The onsets of motor bursts occurring during such episodes
were automatically detected in the integrated traces, and pooled data
according to animal groups (control, longitudinal spinal cord lesion,
etc.) were then transferred to Oriana software (Kovach Computing
Services) for circular phase analysis of the temporal relationship
between activity in selected pairs of muscles or nerves. The results of
this analysis gave the mean vector ␮ and its length r, and two tests
were used to examine the circular distributions: the Rayleigh test (Z),
which tested the concentration of phase values around the mean vector
with a random distribution indicating a lack of coordination between
the two compared burst activities; and the V-test (u), which tested for
a preferential direction (angle) of a given phase distribution [indicated
in the text as u(direction°)]. To simplify graphical representations,
phase values were plotted as grand means of the means of relative
burst onsets throughout individual locomotor episodes. Nevertheless,
the similarity between distributions (for example between control and
lesioned animal groups) was verified for whole populations of events
using the Mardia-Watson-Wheeler test (W). The angular dispersion
around the mean vector ␮ was also calculated to assess the effect of
a given lesion on the power of coupling between different muscle or
nerve discharges, with the size of the angular dispersion value being
inversely proportional to the strength of coupling. A two-tailed nonparametric t-test was used to compare phase dispersions in groups of
control and lesioned animals.
2D kinematics in juvenile animals
Intact and lesioned juvenile frogs were video-taped with a digital
camera (Handycam DCR-PC350E, Sony) while behaving freely in
their standard aquaria or during EMG recording in a smaller experimental tank. No clear evidence was found for a perturbing effect of
the fine EMG wires on the freedom of locomotor movements. The
video files were transferred to a computer using Windows Movie
Maker software (Microsoft) for storage and joint kinematics. The
latter were performed using the free Image J software developed by
W. S. Rasband (U.S. National Institutes of Health, http://rsb.info.nih.
gov/ij/, 1997–2005). Using a manual tracking plug-in supplied by
F. P. Cordelières (Institut Curie), the x and y coordinates of limb joints
J Neurophysiol • VOL
were determined under visual inspection by mouse-clicking on individual frames. Joint angles were determined with Microsoft Excel,
while stick diagrams and angle variations were plotted using CorelDraw 7 (Corel).
RESULTS
Developmental changes in bilateral axial muscle
coordination during swimming in metamorphosing Xenopus
In premetamorphic X. laevis (stages 54 –57), forward locomotion is achieved by rhythmic body undulations produced by
bilaterally alternating spinal ventral root bursts that drive
myotomal muscle contractions in rostrocaudally directed
waves along the tail (Combes et al. 2004; Roberts et al. 1998).
To observe such axial muscle activity in freely behaving
tadpoles, we recorded from the dorsal region of the myotome
of the fifth spinal cord segment, which in the adult corresponds
to the second thoracic segment controlling the back musculature. Consistent with earlier observations from spinal ventral
root recordings in vitro (Combes et al. 2004), EMG recordings
from left and right rostral axial muscles (hereafter referred to as
axial/dorsal (ad)) in intact tadpoles were activated alternately
during episodes of actual locomotion (Fig. 1A). The strongly
clustered distribution of left versus right muscle burst onsets
(n ⫽ 7 episodes; ␮ ⫽ 170.00°; r ⫽ 0.99; Fig. 1A, right) was
further indicative of a strong phase-coupling (Z ⫽ 6.84; P ⬍
0.001) with a phase difference of near 180° [u(180°) ⫽ 3.64;
P ⬍ 0.001], corresponding to symmetrical left-right alternating
contractions necessary for rectilinear propulsion.
During metamorphosis, anurans like Xenopus develop articulated limbs that become functional while the larval tail continues to play an active locomotor role for 1–2 wk prior to its
degeneration (Combes et al. 2004; Nieuwkoop and Faber
1956). During this transitional period, both axial and appendicular locomotor systems can participate conjointly or independently in swimming behavior. As seen in the EMG recordings from a prometamorphic (stages 60 – 61) tadpole in Fig. 1B,
the axial/dorsal muscles display two distinct types of bilateral
coordination according to the mode of locomotion. During
tail-based swimming (n ⫽ 9 episodes; Fig. 1B1), the left and
right ad muscles express a typical larval coordination pattern in
which they are activated alternately [␮ ⫽ 152.11°; r ⫽ 0.93;
Z ⫽ 6.04; P ⬍ 0.001; u(180°) ⫽ 3.07; P ⬍ 0.001], whereas
following a spontaneous switch to limb-based propulsion (12
episodes; Fig. 1B2), the same ad muscles are now active in
synchronous bursts [␮ ⫽ 350.53°; r ⫽ 0.98; Z ⫽ 9.63; P ⬍
0.001; u(0°) ⫽ 4.33; P ⬍ 0.001].
Following metamorphosis (stage 66; 17 animals), the more
rostral axial muscles, which now correspond to postural back
muscles (dorsalis trunci, referred to as dt in Fig. 1C and
following figures), continue to be co-activated during swimming that is now solely limb-based following resorption of the
tail. Thus although uncoordinated bursts may also occur occasionally (not illustrated), circular analysis of left versus right
dorsalis trunci bursts revealed a relatively narrow phase distribution (n ⫽ 37 episodes; ␮ ⫽ 358.59°; r ⫽ 0.98; Z ⫽ 35.39;
P ⬍ 0.001) that had a preferential direction toward phase [0°
(u(0°) ⫽ 8.41; P ⬍ 0.001]. These findings therefore show that
during metamorphosis, dorsal axial muscles alter their pattern
of activation during swimming from strict bilateral alternation
during tail-based locomotion to predominant synchrony during
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
taris longus, which were previously identified as being ankle flexor
and extensor muscles, respectively (d’Avella and Bizzi 2005;
d’Avella et al. 2003). Isolated preparations were then transferred into
a recording chamber and kept under oxygenated frog saline at 16 –
18°C. Suction electrodes were used to record motor activity from left
and right thoracic ventral roots, while petroleum jelly (Vaseline)insulated wire electrodes were used to record bilateral extensor and
flexor activity en passant in distal limb motor nerve branches. Ventral
root and nerve signals were amplified, displayed, and stored in the
same way as for EMG recordings.
All spinal motor output patterns, including those related to swimming, occurred spontaneously without additional chemical or electrical stimulation. Unfortunately, in vitro preparations subjected to
cervico-thoracic or thoraco-lumbar transections were found incapable
of producing bouts of coordinated fictive locomotion, either spontaneously or under classical pharmacological stimulation [with bathapplied N-methyl-D-aspartate (NMDA), bicuculline, serotonin, noradrenalin, either separately or together]. However, as with the intact
spinal cord, isolated preparations with a sagittally lesioned thoracic
cord continued to generate episodes of spontaneous swimming motor
output. Only data from these latter experiments were analyzed and are
illustrated here.
LUMBO-THORACIC COUPLING DURING AMPHIBIAN LOCOMOTION
1375
A
B
C
limb-based propulsion. Moreover, in mixed metamorphic
stages when both tail- and limb-based propulsion occurs, the ad
muscles of a given animal can switch, even within the same
swimming episode, between either coordination pattern according to the type of locomotion being expressed. While the
neural basis of axial muscle recruitment during swimming in
premetamorphic Xenopus tadpoles has been well described
(Roberts et al. 1998), the coordinating mechanisms engaged in
driving these muscles during limb-based locomotion in the
adult are unknown. The following in vivo and in vitro experiments were therefore designed to explore the neural origin of
the synchronous activation of bilateral dorsal trunk muscles
during forward swimming in postmetamorphic animals.
Coordination of hindlimb and dorsal muscles during
rectilinear swimming in juvenile adults
The hindlimb muscle activity patterns that allowed a distinction to be made between rectilinear forward swimming and
turning behavior can be seen in the kinematics analysis and
associated EMG recordings of Fig. 2. Episodes of linear
swimming are characterized by bilaterally symmetrical kick
trajectories of the hind legs (see stick diagrams in Fig. 2A1)
unlike the uncoordinated limb movements associated with
nonlinear swimming (B1). As can be seen in the traces of the
hindlimb knee and ankle joint excursions in Fig. 2, A2 and B2
(top traces), the coordinate movements of the two right limb
joints occurred as an in-phase mirror image of the homologous
left limb joint excursions during a linear swim episode (Fig.
2A2). In contrast, the excursions of these same joints became
asymmetric and irregular during turning maneuvers (Fig. 2B2).
Correspondingly, in simultaneous left-right EMG recordings,
the ankle extensor plantaris longus muscles (referred to as pl in
Fig. 2 and following figures) of the hindlimbs were synchronously active with regularly recurring bursts during rectilinear
swim episodes (Fig. 2A2, bottom traces; see also Fig. 3, B and C),
J Neurophysiol • VOL
whereas the activity of the same muscles became desynchronized during nonlinear swimming (Fig. 2B2, bottom traces). In
the remaining in vivo and in vitro experiments, therefore we
focused attention on episodes of real or fictive swimming in
which synchronous bursting occurred either in the left and right
plantaris muscles or in the corresponding extensor motor
nerves.
To determine how back and hindlimb motor activity is
coordinated during young adult swimming, we made simultaneous EMG recordings in vivo from the left and right plantaris
longus muscles and the third myomeres of dorsalis trunci (Fig.
3, A and B). An initial analysis of the phase relationship
between left and right plantaris longus burst onsets showed a
close coordination (n ⫽ 21 episodes; ␮ ⫽ 4.89°; r ⫽ 0.98; Z ⫽
20.04; P ⬍ 0.001; Fig. 3C, middle polar plot) that was centered
around a 0° phase value [u(0°) ⫽ 6.31; P ⬍ 0.001]. Thus
although some phase dispersion occurred, the main tendency
was for bilateral synchrony to occur between the two extensor
muscles, commensurate with the generation of linear forward
propulsion (see Fig. 2A, compare with B).
In addition to the bilateral synchrony of dorsalis trunci
muscle activity during swimming (Figs. 1C and 3, B and C, left
polar plot), a phase analysis of dorsalis EMG activity versus
ipsilateral plantaris bursts (n ⫽ 23 episodes; ␮ ⫽ 3.59°; r ⫽
0.96; Z ⫽ 21.37; P ⬍ 0.001; Fig. 3C, right polar plot) also
indicated coincident discharge of these trunk and limb muscles
(also see Fig. 3B) with a mean burst onset phase vector that
was directed at 0° [u(0°) ⫽ 6.53; P ⬍ 0.001]. Thus in contrast
to the rostrocaudal propagation of alternating contractions of
the equivalent muscles in the premetamorphic tadpole, the
relative timing of dorsal muscle contractions had altered with
the change in mode of locomotion during metamorphosis, so
that the activation of dorsalis trunci in the froglet was now
phase-locked with bilaterally synchronous hindlimb movements.
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
FIG. 1. Metamorphosis-induced modifications in bilateral
axial/dorsal muscle coordination in free swimming Xenopus
laevis. A: axial muscle coordination in a premetamorphic tadpole (stage 55, left). Raw electromyographic (EMG) recordings
(middle, top 2 traces) and corresponding integrated traces
(bottom 2 traces) from left (L) and right (R) axial/dorsal (ad)
myomeres at the 5th spinal segment level. Right: circular plot
of onset phases of left vs. right axial muscle bursts showing
tendency for alternation (mean vector direction: ⬃180°).
B: coordination of the equivalent muscles in a late-metamorphosing tadpole (stage 60). Same panel arrangement as in A.
Note that depending on the mode of locomotion (shaded area 1
and time base expansion below: larval-like body undulations;
shaded area 2 and time base expansion below: adult-like leg
kicking) the dorsal muscle coordination switched from alternation to synchrony as indicated by the change in mean vector
direction from ⬃180 to ⬃0° in polar plots at right. C: coordination of bilateral dorsalis trunci muscle activity in a postmetamorphic froglet (stage 66). Same panel arrangement as in A.
Synchronous left-right back muscle bursts occur during locomotion that is now exclusively limb-based.
1376
A
A. BEYELER, C. MÉTAIS, D. COMBES, J. SIMMERS, AND D. LE RAY
B
Furthermore, the strength of activation of the dorsal and
hindlimb extensor muscle pairs displayed similar relationships
with swimming speed (mean frequency: 4.4 ⫾ 1.2 Hz; 17
animals). In both cases, area measurements of individual bursts
in the integrated EMG traces revealed a linear and slightly
positive correlation between burst area and the frequency of the
corresponding swim cycle [regression coefficient (rc) ⫽ 0.44,
P ⬍ 0.001, regression slope (a) ⫽ 0.11 and rc ⫽ 0.54, P ⬍
0.001, a ⫽ 0.09 for dorsalis and plantaris muscles, respectively; n ⫽ 450 cycles]. Moreover, the intensities of back
versus limb muscle bursts were themselves significantly and
positively correlated (rc ⫽ 0.24, P ⬍ 0.001, a ⫽ 0.37), which
further pointed to the close temporal relationship between
locomotor-related output produced in the thoracic and lumbar
cord regions.
We next investigated whether this coupling between dorsalis
trunci and plantaris longus activity was mediated by direct
propriospinal interactions between the thoracic and lumbar
cord regions or resulted indirectly from sensory reflexes, as has
been reported to occur, for example, in the maintenance of
human stance (Tokizane et al. 1951). To this end, an in vitro
brain stem/spinal cord preparation, therefore with no movement-related sensory feedback, was used to assess the effects
of deafferentation on the temporal relationships between motor
activity recorded in the second thoracic ventral roots that
normally innervate dorsalis trunci and the distal nerve branches
innervating the ankle extensor and flexor muscles (Fig. 3D). As
previously reported (Combes et al. 2004), such isolated prepJ Neurophysiol • VOL
arations (n ⫽ 7) continue to express spontaneous episodes of
fictive locomotion (Fig. 3E). Although generally shorter (8.2 ⫾
3.5 vs. 10.5 ⫾ 2.9 cycles; P ⬍ 0.05) and slower (1.77 ⫾ 0.5 vs.
4.4 ⫾ 1.2 Hz; P ⬍ 0.001) than swimming episodes in the intact
animal, these in vitro motor patterns continued to exhibit burst
phase relationships that corresponded closely to trunk and limb
movements in vivo (Fig. 3F). Thus homolateral antagonistic
limb nerves were active in alternation (␮ ⫽ 186.06°; r ⫽ 0.64;
Z ⫽ 174.04; P ⬍ 0.001), while homologous bilateral motor
nerves tended to express in-phase bursting (tibialis nerves: ␮ ⫽
26.59°; r ⫽ 0.69; Z ⫽ 134.55; P ⬍ 0.001: plantaris nerves:
␮ ⫽ 343.12°; r ⫽ 0.84; Z ⫽ 190.99; P ⬍ 0.001), in a manner
appropriate for producing alternating cycles of bilaterally synchronous limb extensions and flexions. In addition, burst synchrony was preserved across the isolated cord at the thoracic
segmental level (Fig. 3, E and F: ␮ ⫽ 2.58°; r ⫽ 0.53; Z ⫽
134.55; P ⬍ 0.001) as well as between the thoracic ventral
roots and plantaris nerves originating from the lumbar cord
region (␮ ⫽ 2.22°; r ⫽ 0.63; Z ⫽ 97.97; P ⬍ 0.001). These
results therefore clearly imply that sensory inputs generated by
actual limb and body movements during locomotion are not
essential for maintaining either the coordination of thoracic
motor output to the left and right back muscles or the longitudinal coupling between motor commands to the back and
hindlimb muscles. Nonetheless a significant increase in the
phase dispersion of bilateral thoracic bursts in the isolated
preparation (P ⬍ 0.05; Fig. 3G, compare left histogram pairs)
indicated that sensory information may participate in strength-
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
FIG. 2. Kinematics of free swimming in juvenile adult
Xenopus. A: rectilinear forward swimming in stage 66 froglets.
1: 3 sample images during a complete cycle of hindlimb
extension-flexion with corresponding stick diagrams of back
and hindlimb position reconstructed from markers redrawn on
the images. 2, top: excursions of right and left knee and ankle
angles throughout a 5.5-s forward-swimming episode (- - -,
body axis; 1, 2, the direction of knee excursion during
hindlimb extension). Vertical scale bar: 100°). Bottom traces:
simultaneous EMG recordings from ankle extensor plantaris
longus of the right (R-pl) and left (L-pl) hindlimbs. Note the
strict bilateral synchrony of the muscle bursts. B: nonrectilinear
swimming. Same arrangement as in A. Note the irregularity and
dissociation of limb trajectories (1), knee and ankle excursions
(2, top), and between plantaris longus activity on the 2 sides
(2, bottom).
LUMBO-THORACIC COUPLING DURING AMPHIBIAN LOCOMOTION
A
B
D
E
F
ening the coordination between the left and right dorsalis
muscle contractions. However, other burst phase relationships
(left vs. right plantaris longus and dorsalis vs. plantaris) remained similarly distributed in vitro to their corresponding
muscle EMG patterns in vivo (Fig. 3G, middle and right
histogram pairs).
Effects of spinal lesions on the coordination between
juvenile hindlimb and dorsal muscle activity
To further understand the phase-coupling of back and hindlimb muscle contractions during linear forward swimming, we
made specific lesions to the spinal cords of juvenile frogs (see
METHODS) to define the structural substrate necessary for ensuring such a strong functional relationship.
CERVICO-THORACIC CORD LESIONS. In higher vertebrates including humans, it is commonly acknowledged that the coordination between postural back muscle activity and limb locomotor movements is largely governed by descending influences from supra-spinal centers (Lalonde and Strazielle 2007;
Takakusaki et al. 2004). To assess whether, in the lower
vertebrate Xenopus, dorsalis trunci activation and its coordination with hindlimb movements might also depend on cerebrospinal commands, a series of in vivo and in vitro experiments
was performed in which a complete spinal cord transection was
J Neurophysiol • VOL
made between the last cervical and the first thoracic segments.
Unfortunately, no consistent patterns of fictive swimming
could be evoked in isolated preparations after such a cervicothoracic cord transection (n ⫽ 8; not shown), and only lesions
performed in vivo provided useful data (n ⫽ 11; Fig. 4).
Although these lesioned animals were unable to produce spontaneous locomotor-related activity, tactile stimulation of one of
the hindlimbs could immediately elicit short episodes of linear
swimming kick movements that were similar to those expressed in intact animals (Fig. 4A). Limb and trunk kinematics
during such episodes were also comparable to those of intact
animals (not shown), and again, the left and right plantaris
longus muscles remained active in strict synchrony [n ⫽ 7
episodes; ␮ ⫽ 2.84°; r ⫽ 0.99; Z ⫽ 6.86; P ⬍ 0.001; u(0°) ⫽
3.71; P ⬍ 0.001; Fig. 4B, middle polar plot]. However, the
coupling between these muscles was increased significantly by
the rostral cord transection (W ⫽ 24.68; P ⬍ 0.001) as seen in
the middle histogram pair of Fig. 4C, where the phase dispersion in lesioned animals (■) was compared with equivalent
data obtained from intact control animals (as illustrated in Fig.
3, A–C and G). In cervico-thoracic transected animals, the
activity of left and right dorsalis trunci also remained synchronized [n ⫽ 26 episodes; ␮ ⫽ 5.87°; r ⫽ 0.97; Z ⫽ 24.44; P ⬍
0.001; u(0°) ⫽ 6.95; P ⬍ 0.001; Fig. 4B, left], with the phase
dispersion also being significantly narrower (W ⫽ 34.21; P ⬍
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
FIG. 3. Coordination between bilateral back and hindlimb
motor activity during real and fictive swimming in juveniles.
A–C: analysis of rectilinear forward swimming in vivo.
A: locality of recorded muscles in intact freely behaving animals. Left and right plantaris longus (ankle extensor) and
dorsalis trunci (back muscle) were recorded simultaneously.
B: EMG recordings (top) of plantaris longus (R pl, L pl) and
dorsalis trunci (L dt, R dt) with integrated transforms (bottom)
during free swimming. - - -, the onset of plantaris longus burst
activity. C: circular phase plots of left vs. right back (L/R dt)
and hindlimb (L/R pl) muscle activity and of back vs. leg
muscle activity (dt/pl). Note the tendency for burst synchrony
in all 3 phase relationships. D–F: analysis of fictive swimming
in vitro. D: localization of the somata of motoneurons innervating left (red) and right (green) dorsalis trunci muscles in the
2nd thoracic cord segment (Th2), and isolated brain stem/spinal
cord preparation allowing suction electrode recordings from
dorsalis motor axons in Th2 ventral roots and en passant
recordings from motor nerves to plantaris longus (pl) and
tibialis anterioris (ta). E: simultaneous recordings from bilateral
plantaris (R and L pl) and tibialis (L and R ta) motor nerves
with the left and right ventral roots of the 2nd thoracic segment
(L and R Th2). Corresponding integrated traces are shown
below with - - - positioned at L and R plantaris longus nerve
burst onsets. Such symmetrically organized motor patterns are
presumed to correspond to fictive forward swimming. F: circular phase analysis of burst coordination in vitro, including the
3 relationships corresponding to the in vivo analysis in C (top
3 polar plots). G: comparison of the phase dispersions (in
degrees) in these relationships in vivo (䊐) and in vitro (o). CNS
isolation increased the dispersion of left versus right thoracic/
back phase values (*P ⬍ 0.05), while the distributions of the
two other phase relationships remained unaltered. A–C and
D–F are from 2 different groups of animals.
C
G
1377
1378
A. BEYELER, C. MÉTAIS, D. COMBES, J. SIMMERS, AND D. LE RAY
A
B
FIG. 4. Effect of a complete cervico-thoracic cord transection
on back and hindlimb muscle coordination. A: drawing showing
lesion location and extent, together with sample EMG recordings
and associated integrated traces. B: circular analysis of the 3
indicated burst relationships, corresponding to those analyzed in
intact control animals (see Fig. 3). *** (P ⬍ 0.001) above polar
plots indicate differences (Mardia-Watson-Wheeler test) from
control distributions (as seen in Fig. 3C). C: histograms of phase
dispersions (derived from B) in lesioned animals (■) compared
with intact controls (䊐). ***, significance (P ⬍ 0.001) of differences (2-tailed unpaired t-test) from control group.
C
SAGITTAL LESIONS TO THE THORACIC SPINAL CORD. In premetamorphic Xenopus embryos, the left-right alternation of axial
muscle contractions derives from intra-segmental inhibitory
connections between locomotory circuitry on the two sides of
the cord (Soffe and Roberts 1982; for a review see Roberts
et al. 1998). To assess whether the switch to bilateral syn-
A
E
B
F
C
G
D
H
chrony of dorsalis muscle activation in the adult animal might
result from the establishment of cross-cord excitatory connections during metamorphosis, restricted midline cord sagittal
lesion were made in froglets to separate the two sides of all
three segments of the thoracic cord region. After such a sagittal
lesion (n ⫽ 7; Fig. 5, A–D), animals were still able to express
rectilinear swimming-like behavior (Fig. 5B), and EMG recordings as well as kinematics (not illustrated) did not reveal
any major alterations in the pattern of coordination either
between bilateral hindlimb movements or between the contractions of the left and right dorsalis trunci muscles. Thus the two
plantaris longus muscles were still synchronously active during
swimming episodes [␮ ⫽ 4.28°; r ⫽ 0.80; Z ⫽ 178.57; P ⬍
0.001; u(0°) ⫽ 18.84; P ⬍ 0.001; Fig. 5C] with a coupling
strength that remained similar to intact control animals (W ⫽
2.93; P ⬎ 0.05; and Fig. 5D, middle histogram pair). In
contrast, although the left and right dorsalis trunci also retained
an overall tendency for synchronous activity [␮ ⫽ 14.85°; r ⫽
0.54; Z ⫽ 72.79; P ⬍ 0.001; u(0°) ⫽ 11.66; P ⬍ 0.001; Fig.
5C, left polar plot], the burst onset phases were slightly but
significantly more dispersed than in control (W ⫽ 35.52; P ⬍
0.001; Fig. 5D, left histogram pair). However, a closer inspection of the raw data revealed that this broader distribution
resulted from an increase in the occurrence of supernumerary
FIG. 5. Effect of a sagittal thoracic cord lesion on back
and hindlimb motor burst coordination in vivo and in vitro.
A–D: postlesion analysis of muscle coordination in vivo.
A: schematic of the localization and extent of the midline cord
lesion. F, sites of EMG recordings. B: sample EMG recordings
and corresponding integrated traces. C: circular analysis of
indicated phase relationships (same arrangement as in Fig. 3).
*** (P ⬍ 0.001) and ns (non significant) above polar plots
indicate the significance (Mardia-Watson-Wheeler test) of differences from control distributions (as seen in Fig. 3C).
D: histograms of phase dispersions around the mean vector ␮
for control (䊐) and lesioned (■ corresponding to circular plots in
C) groups of animals. E–H: analysis of left/right and longitudinal motor burst coordination in vitro. Same arrangement as in
A–D with additional recordings (and phase analyses) from left
and right tibialis (L and R ta) motor nerves. *P ⬍ 0.05; **P ⬍
0.01; ***P ⬍ 0.001. A–D and E–H are from 2 different animal
groups.
J Neurophysiol • VOL
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
0.001) than in intact animals (Fig. 4C, left histogram pair).
Similarly, back and leg muscle activity continued to display an
in-phase coordination [n ⫽ 7 episodes; ␮ ⫽ 2.20°; r ⫽ 0.98;
Z ⫽ 5.74; P ⬍ 0.001; u(0°) ⫽ 3.39; P ⬍ 0.001; Fig. 4B, right]
with a burst onset phase distribution significantly different
from control (W ⫽ 17.38; P ⬍ 0.001) due to a significant
decrease in phase dispersion (Fig. 4C, right). As in control
animals, furthermore, the intensity of dorsal and limb extension
muscle bursts remained linearly correlated (rc ⫽ 0.72; P ⬍
0.001; a ⫽ 0.82) after a cervico-thoracic transection. These
data therefore indicated that the cyclic activation of dorsalis
trunci muscles during locomotion was not attributable to descending commands from the brain stem. Rather the actual
strengthening of thoracic cross-cord and lumbo-thoracic coupling following a rostral cord transection was more indicative
of the functional versatility that brain stem influences are
capable of imposing on otherwise inflexible coordination patterns arising from hardwired spinal circuitry.
LUMBO-THORACIC COUPLING DURING AMPHIBIAN LOCOMOTION
TABLE
THORACO-LUMBAR CORD LESIONS. To confirm that the phasic
activation of the left and right dorsalis trunci derives directly
from the lumbar locomotory CPG, complete spinal cord transections were made between the last thoracic and first lumbar
segments. Here again, such lesions prevented locomotorrelated activity in in vitro preparations (n ⫽ 8; not shown), and
meaningful data were only obtained from in vivo experiments
(n ⫽ 11; Fig. 6, A–C) in which the dorsalis or plantaris muscles
were activated by briefly pinching either an anterior or a
posterior limb, respectively. With hindlimb stimulation (Fig.
6A1), only arrhythmic and mostly uncoordinated plantaris
muscle activity was observed. In contrast, EMG-recorded dorsalis (i.e., at a level more rostral to the cord transection)
expressed left and right patterns of activity (Fig. 6, A2 and left
plot in B) in response to a forelimb stimulation (which also
elicited bouts of forelimb movements; not illustrated) that
tended to be synchronous [␮ ⫽ 359.42°; r ⫽ 0.73; Z ⫽ 22.05;
P ⬍ 0.001; u(0°) ⫽ 6.64; P ⬍ 0.001] despite a wider phase
dispersion (W ⫽ 12.08; P ⬍ 0.01; and Fig. 6C, left) than in
intact control animals. As expected, no temporal correlation
was found between dorsalis activity and the occasional plantaris bursts that occurred in response to hindlimb pinching
(␮ ⫽ 302.16°; r ⫽ 0.69; but Z ⫽ 1.40; P ⬎ 0.05; Fig. 6B,
right), further confirming that dorsalis was not being driven via
a possible reflex loop that was being activated by passive
dorsal muscle stretching during hindlimb extensions.
Finally, in two juveniles, the thoracic and lumbar segments
were separated on one side only (left) in combination with a
longitudinal thoracic segment section. Such sagittally/hemicord lesioned animals were still able to swim spontaneously
(Fig. 6D1) or in response to limb tactile stimulation (not
illustrated). Under these conditions, dorsalis activity generally
occurred only on the unlesioned side (Fig. 6D1), and it remained phase-coupled to rhythmic hindlimb extensions [␮ ⫽
334.12°; r ⫽ 0.92; Z ⫽ 6.73; P ⬍ 0.001; u(0°) ⫽ 3.30; P ⬍
0.001; Fig. 6E, right plot]. In contrast, the occasional dorsalis
activity that occurred on the side of the hemi-cord lesion
expressed no clear temporal relationship with hindlimb plantaris contractions (␮ ⫽ 128.89°; r ⫽ 0.73; Z ⫽ 4.31; P ⬍
0.001; nonsignificant V test; Fig. 6E, left plot). Consequently,
the dorsalis versus plantaris phase values for activity on the
disconnected (left) cord side was significantly more dispersed
than for the intact (right) cord side (P ⬍ 0.001; Fig. 6F).
Finally, these bilateral differences were not the result of lesioninduced damage to ipsilateral thoracic motor circuitry since
strong spontaneous bilateral dorsalis activity was occasionally
observed when hindlimb locomotor movements were absent
(Fig. 6D2).
1. Effects of multiple spinal lesions in vivo
Left/right dorsalis trunci (dt)
Left/right plantaris longus (pl)
dt/pl
␮(°)
r
Z
P
u(0°)
P
W
P
10.40
3.36
357.32
0.89 (0.65)
0.99 (0.80)
0.88 (0.59)
7.99
10.81
16.95
⬍0.001
⬍0.001
⬍0.001
3.93
4.64
5.82
⬍0.001
⬍0.001
⬍0.001
3.70
14.53
9.83
⬎0.05
⬍0.001
⬍0.01
Circular statistics of the coordination between left and right dorsalis and plantaris longus muscles, and between back and hindlimb muscles (dt/pl) following
combined cervico-thoracic cord transection, sagittal thoracic section, and thoraco-lumbar rhizotomy. Results of the Rayleigh’s uniformity test (Z), the V-test for
the 0° direction of the mean vector [u(0°)], and the Mardia-Watson-Wheeler test (W) indicate that the three bilateral and longitudinal relationships examined were
not significantly modified by these multiple lesions, although phase distribution variability decreased for I/r plantaris longus and dt/pl coupling. The control r
value for each relationship is indicated in parentheses.
J Neurophysiol • VOL
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
dorsalis activity that did not correspond to hindlimb muscle
contractions (e.g., L dt EMG activity in Fig. 5B). Nevertheless,
these data indicated that excitatory cross-cord pathways within
the thoracic segments were not primarily responsible for the
bilateral synchronization of dorsal muscle activity (see following text).
Significantly, a thoracic cord partitioning did not alter the
linear positive relationship between the relative strengths of
dorsalis and plantaris activation (rc ⫽ 0.56; P ⬍ 0.001, a ⫽
0.37), and their burst firing became more tightly coupled [␮ ⫽
1.88°; r ⫽ 0.82; Z ⫽ 160.23; P ⬍ 0.001; u(0°) ⫽ 17.90; P ⬍
0.001; Fig. 5C right polar plot] than in control animals (W ⫽
17.21; P ⬍ 0.001) as indicated by the reduction in phase
dispersion seen in Fig. 5D (right histogram pair). This stronger
coordination between plantaris and dorsalis activity after a
thoracic sagittal lesion was further confirmed in three additional in vivo experiments where a thoracic midline section
was made in conjunction with both a cervico-thoracic transection and a dorsal rhizotomy of the thoracic and lumbar cord
segments. The overall pattern of lumbo-thoracic activation not
only persisted after such multiple lesions, but they also significantly enhanced the coupling between left versus right plantaris longus and dorsalis versus plantaris muscle activity, as
indicated by the substantial increase in r values for all phase
relationships presented in Table 1. These latter experiments
therefore supported the conclusion that propriospinal projections from the lumbar CPG circuitry to thoracic cord segments are
sufficient and necessary for maintaining the strict lumbo-thoracic
coupling observed during normal rectilinear swimming.
Finally, sagittal thoracic cord lesions to in vitro preparations
(Fig. 5E) had effects on fictive locomotor patterns (n ⫽ 4; Fig.
5, F–H) that were broadly similar to those seen in vivo
(compare with Fig. 5, A–D). This included a persistence of
overall synchrony between left and right thoracic ventral root
bursts (W ⫽ 5.75; P ⬎ 0.05; Fig. 5G, left) despite a significantly increased phase dispersion (P ⬍ 0.05; Fig. 5H, left),
which in turn, and consistent with in vivo lesion experiments
(Fig. 5D, left), implied a residual contribution of thoracic
cross-cord connections to bilateral coupling. More significantly, no changes were observed in the overall coordination
between locomotor-related bursting in the plantaris longus
motor nerves and thoracic ventral roots after a thoracic sagittal
section in vitro (Fig. 5F, compare with recordings from intact
isolated cord in Fig. 3E), although inexplicably, the strength of
the coupling between both left and right limb extensor and
flexor nerve bursts increased significantly (bilateral tibialis
nerves, W ⫽ 54.47; P ⬍ 0.001: plantaris nerves, W ⫽ 78.07;
P ⬍ 0.001) due to a decreased variability in their phase
distributions (Fig. 5H; P ⬍ 0.01 and P ⬍ 0.001, respectively).
1379
1380
A. BEYELER, C. MÉTAIS, D. COMBES, J. SIMMERS, AND D. LE RAY
A
B
C
E
F
DISCUSSION
The fundamental modifications in locomotor behavior that
occur during amphibian metamorphosis (Combes et al. 2004;
Rauscent et al. 2007) require a concomitant adaptation of
postural control and its coordination with locomotion. The
present study explored this changing functional relationship by
recording axial/thoracic motor output to locomotory muscles in
premetamorphic Xenopus tadpoles and after their transition to
postural musculature in postmetamorphic young adults. Associated changes in longitudinal coupling between thoracic and
lumbar cord segments during swimming in juvenile frogs were
also examined through simultaneous recordings from respective back and hindlimb muscles, from their corresponding
motor nerves in isolated nervous system preparations, and after
various spinal cord lesions performed either in vivo or in vitro.
Metamorphosis-induced modifications in
intra- and intersegmental coordination
Whereas in tadpoles the entire axial musculature is recruited
in strict left/right alternation, the dorsal axial muscles controlled by the thoracic region of the young adult spinal cord are
synchronously activated during free forward swimming. Moreover, in contrast to the progressive rostrocaudal activation of
axial muscles in swimming larval Xenopus (Combes et al.
2004; Roberts et al. 1998), back and leg extensor muscles are
co-activated during juvenile hindlimb propulsion. These radical changes in the coordination between left and right thoracic
hemi-segments, and between thoracic and lumbar segments,
therefore implied that the functional reorganization of spinal
motor circuitry that occurs during Xenopus metamorphosis
J Neurophysiol • VOL
(Combes et al. 2004) extends into mid cord regions that are not
involved in producing actual limb movements. Several observations from our in vivo and in vitro experiments provided
insights into the nature of this developmental transformation.
First all spinal lesions in froglets that isolated the thoracolumbar cord from sensory or supra-spinal inputs tended to
narrow the phase distributions of back versus leg muscle
locomotor activity toward even closer synchrony, suggesting
that a preferential functional linkage between the lumbar and
thoracic cord segments is established during the metamorphic
process. Further clear evidence that the dorsal muscle-hindlimb
extensor coupling does not rely on sensory feedback arises
from the observation that the activation patterns of thoracic
ventral roots and leg motor nerves in isolated in vitro preparations remained similar to those of their corresponding target
muscles in vivo. The stronger synchrony observed between
back and hindlimb muscle activity after disconnecting the
thoraco-lumbar cord from the rostral CNS also suggested that
the coupling was not reliant on descending influences from the
brain stem, a situation that differs from rodents and humans
where supra-spinal inputs have been reported to be primarily
responsible for coordinating back and leg muscle contractions
during locomotion (Gramsbergen et al. 1999; Takakusaki et al.
2004). Thus although a remodeling of supra-spinal centers is
undoubtedly required in metamorphosing Xenopus to adapt
descending commands to the changing spinal circuitry, the
altered thoraco-lumbar coordination occurring in young adults
is mainly attributable to de novo synaptic pathways that are
intrinsic to the spinal cord.
Second, that the left/right back muscle synchronization with
bilateral hindlimb extensions persisted after separating the two
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
D
FIG. 6. Effect of thoraco-lumbar cord transections on back
and leg muscle coordination. A–C: complete transection.
A: sample episodes evoked by a brief mechanical stimulation
(Œ) of either a hindlimb (hstim) or forelimb (fstim) after a
thoraco-lumbar transection as indicated in the drawing above.
Left: recordings show uncoordinated dorsal and hindlimb activity during hindlimb movement; right: bilateral dorsalis trunci
activity in the absence of limb muscle activity. B and C: same
arrangement as in Fig. 4 but note the absence of L/R plantaris
longus analysis due to the lack of organized hindlimb motor
activity after such a lesion. **P ⬍ 0.01. D–F: hemi-transection
of left cord side combined with a sagittal lesion of the 3
thoracic segments. D: sample spontaneous episodes recorded
during swimming (1) and in the absence of limb kicking (2).
When the animal swam, the right dorsalis was rhythmically
activated, whereas the left side remained silent. In contrast,
dorsalis bursts could still occur on both sides in the absence of
locomotor activity. E: circular analysis of left (Ldt) and right
(Rdt) dorsalis bursts compared with ipsilateral plantaris (pl)
activity showing the lack of correspondence between left dorsal
and locomotor muscle contractions. F: phase dispersions were
significantly wider (P ⬍ 0.001) on the left, lesioned side.
LUMBO-THORACIC COUPLING DURING AMPHIBIAN LOCOMOTION
A
random after a thoraco-lumbar separation, indicating that coactivating influences on the motor commands to these muscles
may to some extent arise from cerebrospinal inputs and/or the
cervical CPG circuitry responsible for rhythmic forelimb
movements. Indeed it is possible that the occasional participation of forelimb CPG-driven propulsion during hindlimb swimming may also contribute to the coordination of dorsalis
muscle contractions.
Taken together, therefore these findings indicate that descending pathways, sensory inputs and cross-cord connectivity
in the thoracic spinal cord are not essential for coordinating
bilateral back muscle contractions with rhythmic hindlimb
extensions during rectilinear swimming in juvenile Xenopus.
Rather their coordination appears to be determined principally
by ipsilateral projections that ascend from the hindlimb CPG in
lumbar cord segments. Earlier work on spinal cat has also
proposed that the lumbar locomotor CPG may control the
activation sequence of back muscles during walking (Zomlefer
et al. 1984), and in newborn rats, changes in trunk curvature
during locomotion are time-locked with rhythmic hindlimb
stepping, which here also appears to derive from lumbothoracic interactions (Cazalets 2005; Falgairolle and Cazalets
2007).
It is important to remember that the development of limb-based
locomotion in Xenopus is fundamentally different from most other
vertebrates in which effective locomotory behavior depends on
the progressive emergence of functional spinal circuitry from a
nonfunctional embryonic precursor. In Xenopus, an already operational primary locomotor system (for axial-based swimming) is replaced by another during metamorphosis to satisfy
completely different biomechanical requirements (for limbbased propulsion). Figure 7 summarizes the associated developmental changes in spinal network coordination on the basis
of previous knowledge on larval Xenopus (Roberts et al. 1998;
Tunstall and Roberts 1994) and our current findings in young
adult frogs. Essentially spinal connectivity switches from a
system that generates rostrocaudally propagating, bilaterally
alternating motor patterns for undulatory swimming in tadpoles
FIG. 7. Organization of spinal lumbar and thoracic circuitry
for locomotion and posture in preand postmetamorphic Xenopus.
A: in the tadpole, spinal circuitry consists of a distributed chain of
segmental oscillators that generate alternating left-right myotomal
contractions by reciprocal inhibitory cross-cord connections. Propulsive body undulations during swimming are produced by a
rostrocaudal propagation of motor activity that is coordinated by
brain-stem-derived gradients of neuronal excitability extending
progressively down the cord. Schematic derived from earlier work
on postembryonic Xenopus larvae (Tunstall and Roberts 1994;
see also Roberts et al. 1998). B: in the young adult spinal cord,
the dorsal axial musculature is now controlled by weakly
interacting bilateral thoracic circuitry that may be activated by
oppositely directed longitudinal influences: the 1st descends
directly from brain stem postural centers (thin dotted arrows) to
command reactive postural adjustments via appropriate back
and hindlimb muscle activation; the 2nd pathway ascends from
the hindlimb locomotor command network located within the
lumbar cord segments (thick medial arrows). Note that another
pathway, descending from the cervical networks controlling
forelimbs (thin gray medial arrows), may also constitute to
some extent a direct pathway between dorsal motoneurons and
a spinal locomotor network. Although the lumbar CPG requires
activation by brain stem locomotor centers (unfilled lateral
arrows), the lumbo-thoracic pathway enables dynamic postural
control whereby back muscle contractions are matched proactively to ongoing locomotor movements.
B
J Neurophysiol • VOL
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
sides of the thoracic cord suggested that the back muscle
coupling does not rely principally on direct cross-cord connections at the thoracic cord level. This contrasts with the situation
in premetamorphic Xenopus larvae where inhibitory commissural interneurons are responsible for alternating segmental
CPG activity on the two sides of the cord (Roberts et al. 1998).
Evidently, these larval inhibitory interactions are not simply
replaced by equivalent excitatory cross-cord pathways in thoracic segments during metamorphosis to ensure conjoint left/
right dorsalis activation. Rather the persistence of dorsal muscle co-activation after thoracic cord partitioning that remained
coordinated with hindlimb movements indicated that the thoracic intrasegmental coupling was being driven by the lumbar
CPG itself. In direct contrast to the thoracic segments, however, the assembly of hindlimb locomotor circuitry in the
lumbar cord region during metamorphosis must here include
new cross-cord excitatory connections that are necessary for
bilateral limb kick synchrony (see also Combes et al. 2004).
Furthermore, the establishment of longitudinal projections
from the lumbar to thoracic cord regions is also required to
provide the necessary anatomical substrate for coupling dorsalis efferent commands to the hindlimb motor circuitry. While
such ascending propriospinal pathways remain to be identified
in postmetamorphic Xenopus, in Rana pipens, lumbar interneurons have been previously labeled that project directly into
the thoracic spinal cord (Schotland and Tresch 1997). In
Xenopus juveniles, furthermore, these lumbo-thoracic projections appear to be homolateral because a hemicord section
in vivo that disconnected the thoracic circuitry of that side from
the lumbar cord region suppressed activity of the corresponding dorsalis muscle during swimming but did not affect hindlimb kicking-timed activation of the dorsalis on the intact
contralateral side.
Finally, a complete cord transection that separated the thoracic and lumbar segments not only decoupled leg and back
motor outputs during swimming but also resulted in an overall
significant decrease in coincident left-right dorsalis activity.
However, this bilateral relationship never became completely
1381
1382
A. BEYELER, C. MÉTAIS, D. COMBES, J. SIMMERS, AND D. LE RAY
(Fig. 7A) to caudorostrally, bilaterally synchronous activity
that drives back muscle contractions in time with limb-kicking
in the adult (Fig. 7B). In this way, thoracic circuitry, which
contributes actively to body propulsion as an equivalent member of a chain of segmental oscillatory networks in the
premetamorphic larva, becomes subordinate to new, more
caudal, rhythm-generating circuitry that drives hindlimb extension/flexion from the lumbar region of the adult spinal cord.
Posturo-locomotor interactions
J Neurophysiol • VOL
ACKNOWLEDGMENTS
The authors are grateful to M. Falgairolle for helping with kinematics
analysis and R. Nargeot for advice in statistical treatments. We also thank
M.-J. Cabirol-Pol for technical assistance in fluorescence microscopy and L.
Parra-Iglesias for animal care and maintenance. The authors also thank the
European Neuroscience Institute Network (ENI-Net) for valuable discussions.
GRANTS
This work was funded by the entre National de la Recherche Scientifique
(ATIP Jeunes Chercheurs). A. Beyeler was funded by a studentship from the
French Ministère de L’Education Nationale, de l’Enseignement Supérieur et de
la Recherche. The ENI-Net covered part of the publication fees.
REFERENCES
Assaiante C, Malau S, Viel S, Jover M, Schmitz C. Development of postural
control in healthy children: a functional approach. Neural Plast 12: 109 –
118, 2005.
d’Avella A, Bizzi E. Shared and specific muscle synergies in natural motor
behaviors. Proc Natl Acad Sci USA 102: 3076 –3081, 2005.
d’Avella A, Saltiel P, Bizzi E. Combinations of muscle synergies in the
construction of a natural motor behavior. Nat Neurosci 6: 300 –308, 2003.
Ballion B, Morin D, Viala D. Forelimb locomotor generators and quadrupedal
locomotion in the neonatal rat. Eur J Neurosci 14: 1727–1738, 2001.
Bélanger M, Drew T, Provencher J, Rossignol S. A comparison of treadmill
locomotion in adult cats before and after spinal transection. J Neurophysiol
76: 471– 491, 1996.
Beyeler A, Métais C, Combes D, Simmers J, Le Ray D. Propriospinal
coordination of posturo-locomotor function in post-metamorphic Xenopus
laevis. Soc Neurosci Abstr 187.3, 2007.
Cazalets JR. Metachronal propagation of motoneurone burst activation in
isolated spinal cord of newborn rat. J Physiol 568: 583–597, 2005.
Cazalets JR, Bertrand S. Ubiquity of motor networks in the spinal cord of
vertebrates. Brain Res Bull 53: 627– 634, 2000.
Cazalets JR, Borde M, Clarac F. Localization and organization of the central
pattern generator for hindlimb locomotion in newborn rat. J Neurosci 15:
4943– 4951, 1995.
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
For effective displacement during locomotion, coordinated
dynamic postural regulation is required to stabilize body orientation by controlling mainly the head and trunk musculature
(Assaiante et al. 2005; Massion et al. 2004). The interactions
between locomotory and postural systems are generally considered to be based on two separate neural command networks,
both of which are thought to be activated and coordinated
exclusively by supra-spinal commands and sensory feedback
signals (see Mori et al. 2004). Although the postural trunk
muscles of mammals, like in adult Xenopus, are known to
contract rhythmically in time with the step cycle (Falgairolle
and Cazalets 2007; Gramsbergen et al. 1999; Koehler et al.
1984), the neural basis of this coupling has remained unclear,
due undoubtedly to the close functional relationship between
the locomotory and postural systems that renders them difficult
to distinguish during ongoing movement. Furthermore, in most
species including humans, the two systems develop simultaneously and often employ overlapping sets of muscles (Vinay
et al. 2002).
In lower vertebrates such as fish, nonanuran amphibians and
anuran larvae (including Xenopus), locomotion and posture are
achieved by the same sets of axial muscles that are activated
sequentially along the body length during swimming, indicating that the coordination between these two functions occurs at
the individual segmental level (Deliagina et al. 2006). In
postmetamorphic anurans where propulsion is now limb-based,
at least part of the larval dorsal axial musculature has evidently
undergone a functional switch from a mainly locomotory to an
essentially postural role as thoracic trunk muscles in the adult.
Specifically, given its location flanking a rigidified section of
the adult spinal cord, and consistent with a requirement for
dynamic postural control in general (Massion 1994), the most
likely contribution of the third myomere of dorsalis trunci
during locomotion, especially during powerful swimming, is to
increase body stiffness to resist external forces in a similar
manner to epaxial musculature in lizards, for example, which
serves to stabilize the trunk against ground reaction forces
during locomotion (Ritter 1995, 1996).
In lower vertebrates and mammals, both anticipatory and compensatory adjustments to postural perturbations are thought to
depend on supra-spinal influences and sensory feedback from
the periphery (Deliagina et al. 2006; Mori 1987; Mori et al.
2004), thereby implying that spinal postural circuitry serves
merely as a final output stage for higher-order or reflex commands. Our findings add a further dimension to this common
view by indicating that the spinal cord is itself capable of
organizing a major component of postural adjustment by using
intrinsic feed-forward signals from the lumbar pattern-generating networks to predict the consequences of locomotor actions. Interestingly, previous work on cats has shown that
vestibular information is not required for controlling posture,
since responses to postural perturbations are preserved following a bilateral labyrinthectomy (Macpherson and Inglis 1993).
Furthermore, complex motor kinematics and EMG coordination (Bélanger et al. 1996) and body orientation and geometry
(Fung and Macpherson 1999) persist in chronic spinal cats,
indicating here also that supra-spinal commands are not essential for maintaining equilibrium during walking or quiet stance.
However, to our knowledge, a proactive (rather than an anticipatory or reactive) relationship between locomotory movements and postural adjustments that resides solely with propriospinal mechanisms has not previously been clearly established (Stuart 2005).
In conclusion, Xenopus metamorphosis is associated with a
complete reorganization of the postural system from a larval
architecture in which the rostrocaudal recruitment of axial
musculature during swimming implies postural adjustments on
a segment-by-segment basis to an adult en bloc strategy of
postural control where posturo-locomotor coordination extends
over several spinal segments, from the lumbar to the thoracic
cord regions. Furthermore, our findings suggest that Xenopus
offers an attractive model for future studies on the developmental plasticity of posture/locomotion coupling and spinal
network interactions. Especially intriguing in this context is
that both larval and adult motor networks co-exist and can
function separately within the spinal cord at mid-metamorphic
stages of development (see Fig. 1B), thereby providing the
opportunity to explore the neuronal basis of different posturolocomotory strategies within the same organism.
LUMBO-THORACIC COUPLING DURING AMPHIBIAN LOCOMOTION
J Neurophysiol • VOL
Peters SE. Properties of twitch motor units of the ankle extensor muscles in
the bullfrog Rana catesbeiana. J Morphol 221: 121–131, 2005.
Rauscent A, Le Ray D, Cabirol-Pol M-J, Sillar KT, Simmers J, Combes D.
Development and neuromodulation of spinal locomotor networks in the
metamorphosing frog. J Physiol 100: 317–327, 2007.
Ritter D. Epaxial muscle function during locomotion in a lizard (Varanus
salvator) and the proposal of a key innovation in the vertebrate axial
musculoskeletal system. J Exp Biol 198: 2477–2490, 1995.
Ritter D. Axial muscle function during lizard locomotion. J Exp Biol 199:
2499 –2510, 1996.
Roberts A, Soffe SR, Wolf ES, Yoshida M, Zhao FY. Central circuits
controlling locomotion in young frog tadpoles. Ann NY Acad Sci 860:
19 –34, 1998.
Rossignol S, Dubuc R, Gossard JP. Dynamic sensorimotor interactions in
locomotion. Physiol Rev 86: 89 –154, 2006.
Schotland JL, Tresch MC. Segmental and propriospinal projection systems
of frog lumbar interneurons. Exp Brain Res 116: 283–298, 1997.
Shi YB. Amphibian Metamorphosis: From Morphology to Molecular Biology.
New York: Wiley-Liss, 2000.
Soffe SR, Roberts A. Tonic and phasic synaptic input to spinal cord motoneurons during fictive locomotion in frog embryos. J Neurophysiol 48: 1279 –
1288, 1982.
Stuart DG. Integration of posture and movement: contributiuons of Sherrington, Hess, and Bernstein. Hum Mov Sci 24: 621– 643, 2005.
Takakusaki K, Oohinanta-Sugimoto J, Saitoh K, Habaguchi T. Role of
basal ganglia-brainstem systems in the control of postural muscle tone and
locomotion. Prog Brain Res 143: 231–237, 2004.
Thorstensson A, Carlson H, Zomlefer MR, Nilsson J. Lumbar back muscle
activity in relation to trunk movements during locomotion in man. Acta
Physiol Scand 116: 13–20, 1982.
Tokizane T, Murao M, Ogata T, Kondo T. Electromyographic studies on
tonic neck, lumbar and labyrinthine reflexes in normal persons. Jpn
J Physiol 2: 130 –146, 1951.
Tunstall MJ, Roberts A. A longitudinal gradient of synaptic drive in the
spinal cord of Xenopus embryos and its role in coordination of swimming.
J Physiol 474: 393– 405, 1994.
Vallois HV. Les Transformations de la Musculature de L’épisome chez les
Vertébrés. Paris: Doin, 1922.
Vinay L, Brocard F, Clarac F, Norreel JC, Pearlstein E, Pflieger JF.
Develoment of posture and locomotion: an interplay of endogenously
generated activities and neurotrophic actions by descending pathways. Brain
Res Rev 40: 118 –129, 2002.
Zomlefer MR, Provencher J, Blanchette G, Rossignol S. Electromyographic
study of lumbar back muscles during locomotion in acute high decerebrate
and low spinal cats. Brain Res 290: 249 –260, 1984.
100 • SEPTEMBER 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 18, 2017
Combes D, Merrywest SD, Simmers J, Sillar KT. Developmental segregation of spinal networks driving axial- and hindlimb-based locomotion in
metamorphosing Xenopus laevis. J Physiol 559: 17–24, 2004.
Deliagina TG, Orlovsky GN, Zelenin PV, Beloozerova IN. Neural bases of
postural control. Physiology 21: 216 –225, 2006.
Falgairolle M, Cazalets JR. Metachronal coupling between spinal neuronal
networks during locomotor activity in newborn rat. J Physiol 580: 87–102,
2007.
Fung J, Macpherson JM. Attributes of quiet stance in the chronic spinal cat.
J Neurophysiol 82: 3056 –3065, 1999.
Gramsbergen A, Geisler HC, Taekema H, van Eykern LA. The activation
of back muscles during locomotion in the developing rat. Dev Brain Res
112: 217–228, 1999.
Grillner S. Control of locomotion in bipeds, tetrapods and fish. In: Handbook
of Physiology. The Nervous System. Motor Control. Bethesda, MD: Am.
Physiol. Soc., 1981, sect. 1, vol. II, part 2, p. 1179 –1236.
Grillner S. Biological pattern generation: the cellular and computational logic
of networks in motion. Neuron 52: 751–766, 2006.
Juvin L, Simmers J, Morin D. Propriospinal circuitry underlying interlimb
coordination in mammalian quadrupedal locomotion. J Neurosci 25: 6025–
6035, 2005.
Kiehn O. Locomotor circuits in the mammalian spinal cord. Annu Rev
Neurosci 29: 279 –306, 2006.
Koehler WJ, Schomburg ED, Steffens H. Phasic modulation of trunk muscle
efferents during fictive spinal locomotion in cats. J Physiol 353: 187–197, 1984.
Lalonde R, Strazielle C. Brain regions and genes affecting postural control.
Prog Neurobiol 81: 45– 60, 2007.
Langlet C, Leblond H, Rossignol S. Mid-lumbar segments are needed for the
expression of locomotion in chronic spinal cats. J Neurophysiol 93: 2474 –
2488, 2005.
Macpherson JM, Inglis JT. Stance and balance following bilateral labyrinthectomy. Prog Brain Res 97: 219 –228, 1993.
Massion J. Postural control system. Curr Opin Neurobiol 4: 877– 887, 1994.
Massion J, Alexandrov A, Frolov A. Why and how are posture and movement coordinated? Prog Brain Res 143: 13–27, 2004.
Mori S. Integration of posture and locomotion in acute decerebrate cats and in
awake, freely moving cats. Prog Neurobiol 28: 161–195, 1987.
Mori F, Nakajima K, Tachibana A, Takasu C, Mori M, Tsujimoto T,
Tsukada H, Mori S. Reactive and anticipatory control of posture and
bipedal locomotion in a nonhuman primate. Prog Brain Res 143: 191–198,
2004.
Nieuwkoop PD, Faber J. Normal Table of Xenopus laevis (DAUDIN). A
Systematical and Chronological Survey of the Development From the
Fertilized Egg Till the End. Amsterdam: North Holland, 1956.
1383